Prof Willem Fourie, lead of the Policy Innovation Lab, spoke at an international conference on sustainability cooperation in Bonn, Germany, in June, setting out how artificial intelligence bears on the future of global development goals. His argument was that access to AI could increasingly be treated as a matter of social policy.
The conference, Sustainability Cooperation Beyond 2030, was hosted by the German Institute of Development and Sustainability (IDOS) and brought together researchers and practitioners from around the world, ahead of formal international discussions due at the Sustainable Development Goal Summit in 2027. Fourie was invited to address a theme largely absent from the original 2030 agenda, namely how AI is reshaping the prospects for international cooperation, and did so in a session on the social dimension of sustainability.
He posed a hypothetical question: what if access to capable language models comes to resemble access to a basic utility such as electricity? There are obvious differences between language models and utilities, but their pervasive impact across all dimensions of social life is increasingly similar. On this view, guaranteeing equitable and uninterrupted access becomes a concern for national governments, in the way that utility provision usually is. The argument raises at least three issues.
The first is access itself. Like other utilities, AI depends on access to more basic utilities, notably reliable electricity and connectivity, so its distribution cannot be separated from existing infrastructure gaps. This brings a set of governance questions into focus: who has reliable access, under what conditions, with what safeguards, and whether unequal access risks widening existing inequalities rather than reducing them.
The second is sovereignty. The utility framing places responsibility for access with national governments, but here the analogy also reaches its limits. Most countries can generate their own electricity, but few train or host advanced language models. Provision therefore sits largely beyond national borders, which raises the risk that access could be reduced or withdrawn by outside entities, and the more deeply these systems are integrated, the more consequential that dependence becomes. It also means that equitable and uninterrupted access cannot be safeguarded at the national level alone. For many countries it can only be secured through regional or global cooperation.
The third is that AI differs from electricity in kind, not only in how it is provided. Its potential benefits could be transformational, but it also carries risks that a basic utility does not, including the deskilling of workforces and the erosion of institutional capabilities.
In line with the Policy Innovation Lab’s emphasis on responsible AI for transformative policymaking, the Lab will continue to contribute to national and global conversations on appropriate safeguards and governance mechanisms, at national, regional and global levels.
This article was drafted using human expertise supported by AI-assisted writing tools.